1
|
|
|
package main |
2
|
|
|
|
3
|
|
|
import ( |
4
|
|
|
"encoding/xml" |
5
|
|
|
"fmt" |
6
|
|
|
"time" |
7
|
|
|
"net/http" |
8
|
|
|
"os" |
9
|
|
|
) |
10
|
|
|
|
11
|
|
|
// URLSet is root for site mite |
12
|
|
|
type URLSet struct { |
13
|
|
|
XMLName xml.Name `xml:"urlset"` |
14
|
|
|
XMLNs string `xml:"xmlns,attr"` |
15
|
|
|
URL []URL `xml:"url"` |
16
|
|
|
} |
17
|
|
|
|
18
|
|
|
// URL is for every single location url |
19
|
|
|
type URL struct { |
20
|
|
|
Loc string `xml:"loc"` |
21
|
|
|
LastMod string `xml:"lastmod,omitempty"` |
22
|
|
|
ChangeFreq string `xml:"changefreq,omitempty"` |
23
|
|
|
Priority float32 `xml:"priority,omitempty"` |
24
|
|
|
} |
25
|
|
|
|
26
|
|
|
func (us *URLSet) saveToFile(filename string) error { |
27
|
|
|
m, err := xml.Marshal((*us)) |
28
|
|
|
if err != nil { |
29
|
|
|
return err |
30
|
|
|
} |
31
|
|
|
|
32
|
|
|
file, err := os.OpenFile(filename, os.O_WRONLY|os.O_CREATE|os.O_TRUNC, 0777) |
33
|
|
|
file.Write([]byte(xml.Header)) |
34
|
|
|
file.Write(m) |
35
|
|
|
file.Close() |
36
|
|
|
return err |
37
|
|
|
} |
38
|
|
|
|
39
|
|
|
func (us *URLSet) validate() URLSet { |
40
|
|
|
client := &http.Client{ |
41
|
|
|
Timeout: 10*time.Second, |
42
|
|
|
} |
43
|
|
|
c := make(chan string, 20) |
44
|
|
|
|
45
|
|
|
validURLs := []URL{} |
46
|
|
|
for _, url := range (*us).URL { |
47
|
|
|
go func(url URL, c chan string) { |
48
|
|
|
resp, err := client.Get(url.Loc) |
49
|
|
|
defer func() { <-c }() |
50
|
|
|
if err != nil { |
51
|
|
|
c <- err.Error() |
52
|
|
|
return |
53
|
|
|
} |
54
|
|
|
c <- fmt.Sprintf("Response code is %d for %s", resp.StatusCode, url.Loc) |
55
|
|
|
if resp.StatusCode == 200 { |
56
|
|
|
validURLs = append(validURLs, url) |
57
|
|
|
} |
58
|
|
|
}(url, c) |
59
|
|
|
} |
60
|
|
|
|
61
|
|
|
for range us.URL { |
62
|
|
|
fmt.Println(<-c) |
63
|
|
|
} |
64
|
|
|
newURLSet := URLSet{ |
65
|
|
|
XMLNs: us.XMLNs, |
66
|
|
|
} |
67
|
|
|
for _, url := range validURLs { |
68
|
|
|
newURLSet.URL = append(newURLSet.URL, url) |
69
|
|
|
} |
70
|
|
|
return newURLSet |
71
|
|
|
} |
72
|
|
|
|
73
|
|
|
func newURLSetFromXML(rawXMLData []byte) URLSet { |
74
|
|
|
us := URLSet{} |
75
|
|
|
|
76
|
|
|
err := xml.Unmarshal(rawXMLData, &us) |
77
|
|
|
|
78
|
|
|
if err != nil { |
79
|
|
|
fmt.Printf("Sitemap cannot parsed. Because: %s", err) |
80
|
|
|
return URLSet{} |
81
|
|
|
} |
82
|
|
|
return us |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
func singleProcess(uri string, filename string) { |
86
|
|
|
client := &http.Client{ |
87
|
|
|
Timeout: 10*time.Second, |
88
|
|
|
} |
89
|
|
|
|
90
|
|
|
if Verbose {fmt.Printf("Single process started for %s\n",filename)} |
91
|
|
|
resp, err := client.Get(uri) |
92
|
|
|
if err != nil { |
93
|
|
|
fmt.Printf("Url cannot fetched: %s\n", uri) |
94
|
|
|
fmt.Println(err) |
95
|
|
|
os.Exit(1) |
96
|
|
|
} |
97
|
|
|
|
98
|
|
|
rawXMLData := readXMLFromResponse(resp) |
99
|
|
|
|
100
|
|
|
urlSet := newURLSetFromXML(rawXMLData) |
101
|
|
|
if Verbose {fmt.Printf("URLSet Generated.\n")} |
102
|
|
|
|
103
|
|
|
newURLSet := urlSet.validate() |
104
|
|
|
if Verbose {fmt.Printf("URLSet Validated.\n")} |
105
|
|
|
|
106
|
|
|
err = newURLSet.saveToFile(filename) |
107
|
|
|
|
108
|
|
|
if err != nil { |
109
|
|
|
fmt.Println(err) |
110
|
|
|
os.Exit(1) |
111
|
|
|
} |
112
|
|
|
} |
113
|
|
|
|